SUMMARY: keyserv is dying

From: Nadya Williams (nadya@rai.ucsd.edu)
Date: Thu Jun 03 1999 - 12:52:37 CDT


Thanks to Marina Daniels and Eirh-Yu Hsie.

SUMMARY:
=========
Both pointed out that the patch 105568-13 can cause a keyserv to coredump.
Marina kindly provided the whole summary (one I could not find) from a
previous posting.

FIX:
====
The fix is to backout the patch, and go to a previous patch edition.
I installed 105568-14, hopefully this will fix the problem. Its just that the
README file for that patch doesn't say anywhere that it will fix a keyserv
bug...

If this one fails, will go back to patch revision 12.

Original Question:
==================
After installing the latest 2.6 Recommended patch cluster I have a strange
problem with the keyserv. It dies for no apparent reason, and thus makes
login, sendmail (for pop accounts) and other services that require passwd
unavailable. I rebooted the server when it happened the first time, then it
happened again after about 2-3 days of running just fine.

1. System: Ultra-30, SunOS 5.6, latest patches applied, NIS+ master.
   The replica, and clients that had the same patches applied have no problems.

2. I tried to debug the core file with adb, and it looks like the culprit is a
   door_server() which corrupts the stack. here is the snip:
   tstate: 882001a02 (ccr=0x8, asi=0x82, pstate=0x1a, cwp=0x2)
   pstate: ag:0 ie:1 priv:0 am:1 pef:1 mm:0 tle:0 cle:0 mg:0 ig:0
   pc 0xef6c245c door_server+0x38: ld [%i4 + 0xe00], %i3
   npc 0xef6c248c door_server+0x68: call 0xef701d9c

   The door_server() lives in libnsl.so, which was updated by the 105401-20
   patch in the latest cluster.
   
3. All I get from dbx (since no symtable is available) is :
        t@0 (l@3) terminated by signal SEGV (Segmentation Fault)
        dbx: core file read error: address 0xef651de8 not in data space
        dbx: core file read error: address 0xef651e24 not in data space
   The first read error corresponds to the stack pointer that adb gives.

I am using a temporary solution now which is running a cron job that checks if
the keyserv is running, and if not restarts it. However, I would like to get to
the root of the problem, and fix it properly.
---------------------------------------------------------------------------

Nadya Williams
nwilliams@ucsd.edu



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:13:20 CDT